OpenAI has announced that it is expanding the rollout of its Advanced Voice Mode (AVM) feature to more ChatGPT users, specifically targeting those subscribed to the Plus and Teams plans. This move will provide a more conversational experience with ChatGPT and is expected to extend to Enterprise and Education tier customers starting next week.
As part of this wider release, AVM has received a visual redesign. Instead of the original animated black dots showcased back in May, the feature now appears as a blue, animated sphere. When AVM becomes available, users will be alerted with a pop-up notification next to the voice icon within the ChatGPT app.
Along with this update, ChatGPT is introducing five new voices: Arbor, Maple, Sol, Spruce, and Vale. This addition expands the voice options to a total of nine, matching closely with Google’s Gemini Live, which offers nearly the same number. The new voices join existing ones—Breeze, Juniper, Cove, and Ember—all of which have nature-inspired names, aligning with AVM’s objective to deliver a more organic and natural-sounding interaction.
Notably missing from the lineup is the “Sky” voice, which was initially demonstrated during OpenAI’s spring update but later removed due to legal challenges from actress Scarlett Johansson. Johansson claimed that the voice bore too close a resemblance to her own, reminiscent of her role as an AI system in the movie “Her.” OpenAI promptly discontinued the use of Sky, asserting that it was never intended to mimic the actress’s voice, despite references to the film made by some OpenAI staff on social media.
Another feature absent from this rollout is the video and screen-sharing functionality that OpenAI introduced during its spring showcase. This multimodal capability would have enabled GPT-4 to process both visual and audio inputs simultaneously. During the initial demo, an OpenAI team member demonstrated how ChatGPT could respond to real-time questions about handwritten math problems or code displayed on a computer screen. However, OpenAI has not yet provided any timeline for when these multimodal features might become available.
OpenAI claims it has made enhancements to AVM since its alpha testing phase, particularly in terms of understanding accents more accurately and facilitating smoother, faster conversations. Although earlier tests revealed occasional glitches, the company insists that these issues have since been addressed.
Furthermore, customization features like Custom Instructions, which allow users to tailor how ChatGPT interacts with them, and the Memory feature, which lets ChatGPT retain conversational context for future reference, are also being integrated into AVM. However, OpenAI has confirmed that AVM is currently unavailable in certain regions, including the European Union, the U.K., Switzerland, Iceland, Norway, and Liechtenstein.
